NTS: Authenticated Time at Meta
engineering.fb.com> Today it is load-bearing:
Please no.
You're absolutely right to question it. And here is where it gets interesting -- they probably talked to Claude enough to start talking like Claude without even realizing it.
Can Hacker News PLEASE add a no-human filter? Reads as AI on first pass, but missing proper em dash names the author. Verdict: dead giveaway.
For fucks sake this reeks of being written by an LLM, but not the usual Claude or ChatGPT vibe - probably Meta's own AI. Endearing to see it's also going for the "load-bearing" trap.
Usually the one that writes the code writes the blog post.
> ... or whenever a correction lands
> The protocol is finished, the servers are appearing, the client implementations exist.
Besides "load-bearing", I also do not understand the fondness of LLMs for "lands".
While the LLM verbal tics are annoying, the content of this Meta article, which is about the newer NTP-related RFCs, RFC 8915 and RFC 9769, and also the Meta open-source time repository, are useful.
> While the LLM verbal tics are annoying, the content of this Meta article, which is about the newer NTP-related RFCs, RFC 8915 and RFC 9769, and also the Meta open-source time repository, are useful.
The problem is, I have a hard time distilling content out of the LLM fluff.
Some guy: "Muse, have a Claude 4x model write up the post. Sound human".
Quarterly meeting: "Internal Muse adoption has been skyrocketing and really being a force multiplier in..."
Just a lot of slop juice...
Utter AI bollocks.